The Locations of Word Segmentation in Chinese Reading: Research Based on the Eye-Movement-Contingent Display Technique

Abstract

Recent studies have demonstrated that words are fundamental units in Chinese reading (Bai et al., 2008; Yan et al., 2010). Yan et al. (2010) suggested a two-stage process model for Chinese reading. If parafoveal word segmentation is successful for target selection, readers will aim at the center of the target word. If not, readers will gaze at the beginning of the next word. Given these hypotheses, we examine whether the assistance in word segmentation promotes reading. Two experiments were conducted using the eye-movement-contingent display technique. The sentences used in both experiments consisted of 7 to 10 two-character words. These stimuli were balanced following a Latin-square design. There were four treatments of sentences in Experiment 1. As Word n was fixated, (1) the color of Word n and all words to its left changed from red to black, (2) the color of Word n+1 and all words to its left changed from red to black, (3) only the color of Word n changed from red to black, and (4) normal sentences were presented in red. Moreover, the four corresponding treatments from black to red were added to correspond to these changes. The first treatment facilitated the segmenting of Word n from its subsequent text, the second facilitated the segmenting of Word n+1, the third drew attention to Word n, and the normal sentences provided a baseline. As a result, there were no significant differences in reading time and the number of fixations among four experimental treatments (p0.05). However, the mean gaze time and the number of words refixated in the second treatment were lower than those in the baseline group (p0.05), whereas the mean saccade length and the number of words skipped were higher (p0.05). The results of Experiment 1 showed that Chinese readers always segmented Word n+1 from its subsequent texts as they fixated on Word n. However, we cannot exclude the possibility that exogenous attention leads readers to fixate for less time. Therefore, a second experiment was conducted. Experiment 2 adopted similar treatments as Experiment 1, but the two adjacent characters not belonging to a word were grouped and changed color together. In other words, the manipulations did not provide cues to facilitate word segmentation. Consequently, there were no significant differences in mean saccade length, the number of fixations, the number of words refixated and skipped, and the number of regressions among different treatments (p0.05). However, there were reliable differences in mean fixation duration, gaze duration, first fixation duration, the number of words skipped and sentence reading time between the third treatment and the baseline group (p0.05), indicating that the treatment significantly interrupted reading. The patterns of eye movements in Experiment 2 differed from those in Experiment 1. Thus, the results obtained in Experiment 1 were not due to exogenous attention. The distribution of the first fixation on the word zones was checked, and the fixation duration in the single-fixation situation and the first fixation duration in the two-fixation situation were compared. The latter was longer than the former in Experiment 2, whereas there was no difference in Experiment 1. Therefore, we concluded that there were two types of word segmentations. First, readers attempted to segment Word n+1 from its subsequent text as they fixated on Word n. Second, readers continued to segment the word as it was fixated on if the first segmentation failed.